Confidence Intervals for Estimated Saturated Hydraulic Conductivity Measured Using Compact Constant Head Permeameters.

How many field measurements are necessary to meet a pre-determined confidence level when measuring in situ saturated hydraulic conductivity (K_sat)?  For any given landscape there are a minimum number of samples or replications needed to accurately estimate the true mean K_sat within an areal unit of soil.  Fewer replications than the minimum threshold could fail to meet a desired confidence level.  Conversely, resources invested in the collection of unnecessary observations deprive additional locations of study.  The objective of this study is to determine the threshold number of samples necessary and at what spacing they should be in order to determine specific confidence intervals for estimating the K_sat.  For this study, two selected sites near Morgantown, WV in Major Land Resource Area 126 (Central Allegheny Plateau) were selected.  Both sites occur in undifferentiated soil map units of Culleoka-Westmoreland silt loams (fine-loamy, mixed, active, mesic Ultic Hapludalfs), 8 to 15 percent slopes.  The major difference between Culleoka and Westmoreland is the depth to bedrock (moderately deep vs. deep, respectively).  At each site K_sat measurements were taken via a compact constant head permeameter (Amoozemeter).  A single measurement was taken in the upper 16 to 20 cm of the Bt horizon at each point in a triangular 13 m grid across each selected sub-map unit delineation.  The variability within the data was used to determine the minimum distance needed between samples to avoid autocorrelation within the data.  Random subsets of the data were used to calculate mean K_sat values as well as to establish the frequency at which the subset means and the overall mean differ significantly.  Utilizing these narrowly defined conditions has allowed us to identify the number and spacing of samples needed to measure K_sat at pre-determined confidence levels.